maker-space / structured_experiment_chain.py
isayahc's picture
decoupling and differentiating the 2 different chains
69194ec
raw
history blame
4.43 kB
"""
#TODO: make a agent that uses HUMAMN as a tool to get:
- Purpose of science experiment
- What fields of study do they already know of
#IDEA: Platform generate more indepth experiments by generaing a data set and generate / collect scienfic data
### Chatbot
the chatbot helps the BOUNTY_BOARD_CHAIN generate science experiments
### EXPERIMENT and Provide feedback on experiments
### Interrgration
- I need to intergrate this code into the app. This includes creating an id for each post, and potentially and a comment section for each "Experiment"
- I addition i need to generate a mostly pinecone retriever to geenrate scientific experiments from the "community vectore search"
- potentially have prenium users store their private data, but i may not implement this during the hackathon
"""
# https://python.langchain.com/docs/modules/model_io/output_parsers/types/structured
from langchain.output_parsers import ResponseSchema, StructuredOutputParser
from langchain.prompts import PromptTemplate
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain.memory import ConversationBufferMemory
from langchain_core.runnables import RunnablePassthrough
from langchain.retrievers import ArxivRetriever, pubmed
from langchain_core.output_parsers import StrOutputParser
from langchain.retrievers import ArxivRetriever
from langchain.retrievers import PubMedRetriever
from langchain.retrievers import WikipediaRetriever
from operator import itemgetter
response_schemas = [
ResponseSchema(name="Experiment_Name", description="the name given to the experiment"),
ResponseSchema(name="Material", description="list of materials need to perfrom the experiments", type="list"),
ResponseSchema(name="Sources", description="list of sources where the information was retrievered from", type="list"),
ResponseSchema(name="Protocal", description="detailed instructions On how to make the item or perform the experiment", type="list"),
ResponseSchema(name="Fields_of_study", description="the fields of study that his experiment uses", type="list"),
ResponseSchema(name="Purpose_of_Experiments", description="assume what the user is trying to acchieve"),
ResponseSchema(name="Safety_Precuation", description="What does the User need to know to avoid any potential harm"),
ResponseSchema(name="Level_of_Difficulty", description="How difficult is it to perform this experiment ecample beginner, novice, Intermidiate, Hard "),
#
]
output_parser = StructuredOutputParser.from_response_schemas(response_schemas)
memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
format_instructions = output_parser.get_format_instructions()
prompt = maker_prompt = PromptTemplate(
template="You must generate a well detailed list of items for creating a given item from scratch. \
Also describe the purpose for a text-to-3d model to use for extra context\n{format_instructions}\n{question}\n{context}",
input_variables=["question"],
partial_variables={"format_instructions": format_instructions},
memory = memory
)
def join_strings(*args: str) -> str:
"""
Join an arbitrary number of strings into one string.
Args:
*args: Variable number of strings to join.
Returns:
str: Joined string.
"""
return ''.join(args)
def format_docs(docs):
return "\n\n".join([join_strings(d.page_content, d.metadata['Entry ID'],d.metadata['Title'], ) for d in docs])
arxiv_retriever = ArxivRetriever(load_max_docs=2)
# model = ChatOpenAI(temperature=0)
model = ChatOpenAI(temperature=0,model="gpt-4")
retriver = arxiv_retriever = ArxivRetriever(load_max_docs=2)
pub_med_retriever = PubMedRetriever()
wikipedia_retriever = WikipediaRetriever()
arxiv_chain = (
{"context": arxiv_retriever, "question": RunnablePassthrough()}
| prompt
| model
| output_parser
)
pub_med_chain = (
{"context": pub_med_retriever, "question": RunnablePassthrough()}
| prompt
| model
| output_parser
)
wikipedia_chain = (
{"context": wikipedia_retriever, "question": RunnablePassthrough()}
| prompt
| model
| output_parser
)
if __name__ == "__main__":
query = "how to create alectronoic on a cellulose subtstrate"
pub_med_data = pub_med_chain.invoke(query)
wiki_data = wikipedia_chain.invoke(query)
x=0